The Sunday Star Times has a story about the latest Horizon poll, which shows that, while the Nats are comfortably ahead of Labour, the difference between the left and right is small.
This is good news for the left for sure, even if the polling methodology Horizon uses suggests that the results may not be as credible as those of the polls that use phone polling.
From what I can tell anyone can register to vote on the Horizon political poll, so it would be quite easy for a small group of committed activists to influence poll results. I'm not saying that has happened, but it is noticeable the trend in the Horizon Poll is for the left parties to register higher support than in the other regular polls (TV3, TVNZ, Roy Morgan). The other regular polls indicate that the gap between left and right is still quite big.
Phone polling has its own problems, because a lot of people with limited means don't have landlines, and this may skew some results towards the right parties. But whether this is a major factor or not isn't clear.
It's also pretty obvious that the two TV polls are the only ones that count, at least in the minds of the mainstream commentariat. A day after the Roy Morgan poll came out showing a lift in support for Labour and the Greens, I was still hearing people on the radio talking about how damning of Labour's tax package the TVNZ poll was.
All of this suggests that, while polls are useful measures of where public support for political parties resides at any given time, we should be wary of taking too much from them.
Sunday, July 24, 2011
2 comments:
I welcome comments, but I ask commenters to follow a few simple rules:
1. I delete anonymous comments. Please use either a name or moniker. I am not asking anyone to reveal their secret identity. Just don't call yourself "Anonymous".
2. Please don't abuse or defame others.
3. Moronic or nonsensical comments may be deleted.
4. I don't often exercise the heavy hand of censorship, but I do reserve the right to delete any comment I don't like, for any reason.
Subscribe to:
Post Comments (Atom)
Horizon's polls may be the most accurate.
ReplyDeleteAt least they give a better picture of the whole political marketplace and will be a better indicator of the trends before the election according to this political economist http://www.horizonpoll.co.nz/page/135/political-sc
Why on earth do the media companies continue to publish polls which do not determine if a respondents are eligible to vote, ask them if they are going to vote; and fail to ask the undecided what their preferences are? Instead they just publish decided results as a percentage of 100. Hence the Herald DigiPoll confessed to this a fortnight ago, and admitted there were 11.9% undecided. NO party could have over 51% if this total market view is taken. Same with the other polls.
What's more Horizon weights its samples on six factors, including party vote 2008. It says this can affect the results by up to 5%:http://www.horizonpoll.co.nz/page/133/political-po
It has even recast a Digipoll to show how its results and Horizon's come wthin their margisn of error if this is done.
So why do we put up with poll results with small samples which exclude 30% of the market - and pretend to tell us how the minor parties are faring, or even what the coalition possibilities are?
Thank goodness Horizon is breaking new ground here - and with a very robust sample selection and weighting system.
It says, for those who are serious about research methods, that while it is possible to join its HorizonPoll main panel, more than 90% are recruited to specfically match the population. So it's not self selecting. The weighting system then ensures respondent samples are again balanced to reflect the national population.
Thanks goodness for an intelligent use of new technology.
Anon, out of interest, how many shares in Horizon do you own?
ReplyDelete